Sophont

What Sophont does

Sophont is an AI-native medical research lab building multimodal foundation models that unify clinical notes, laboratory results, medical imaging (X-ray, CT, MRI), pathology slides, ECG/wearable signals, genomic data, medications, and demographics into a single patient state representation. Founded in 2025 by Tanishq Mathew Abraham (former Research Director at Stability AI, UC Davis PhD) and Paul Scotti (former Head of NeuroAI at Stability AI, Princeton postdoc), the company maintains an open-source-first strategy with publicly released models including OpenMidnight (pathology foundation model trained for under $1,600) and CortexMAE (functional MRI foundation model), alongside evaluation benchmarks Medmarks (medical LLM) and Brainmarks (fMRI). The company anchors MedARC, described as the world's largest open medical AI research community with 4,000+ members, which functions as both a research collaboration platform and a developer adoption funnel.

The company's research output is substantial for its stage: 10 peer-reviewed publications across NeurIPS Spotlight, ICML, CVPR, and Nature Biomedical Engineering, with applications spanning disease trajectory prediction, intervention response modeling, biomarker discovery, and fMRI-to-image reconstruction. The team also includes Hassan Muhammad (PhD, Weill Cornell; formerly MSKCC Research Scientist and Senior Director of AI at Pathomiq) as Head of Applied Research.

Sophont's business model is dual-track: an open-source distribution channel (free models and benchmarks released publicly) layered with an enterprise commercialization motion. The first commercial product is in active development with unnamed clinical, biopharma, and life sciences partners, with planned revenue from enterprise SaaS subscriptions and usage-based API access. Pricing is quote-based with no public figures disclosed. The company has raised $9.22M in seed funding (September 2025) led by Kindred Ventures with participation from Delphi Ventures, Upfront Ventures, Factorial Capital, Headwater Ventures, AICONIC Ventures, Brightwing Capital, and angel investors including Google Chief Scientist Jeff Dean, Google DeepMind's Logan Kilpatrick, Hugging Face CEO Clem Delangue, and Weights & Biases co-founder Lukas Biewald. The company is headquartered in Sacramento, California, and operates with a small team (1–10 employees).

Name

Sophont

Legal name

Sophont

Website

https://sophontai.com

Company type

Private

Founded year

2025

Operating status

Operating

Headcount range

1–10 employees

Short description

Sophont builds open-source multimodal foundation models for medicine, unifying clinical notes, imaging, pathology, genomics, and signals into unified patient representations. It serves biopharma, hospitals, and life sciences researchers, anchoring the 4,000+ member MedARC open medical AI community.

Ownership category

akta.pro rank

Product category

Medical AI Foundation Models

NAICS

Research and Development in Biotechnology (except Nanobiotechnology) (541714)

SIC

Services-Commercial Physical & Biological Research (8731)

akta.pro industry

PRIMARYHLAIALAF

Translational Medicine & Biomarkers

Healthcare & Life Sciences/Pharmaceuticals/Pharmaceutical Research and Development

SECONDARYHLAGANAI

Bioinformatics, Multi-Omics Data Analysis & Interpretation Services

Healthcare & Life Sciences/Life Science Tools & Services (CRO/CDMO/Lab Services)/Specialty Lab Services (Genomics/Proteomics/Cellomics)

SECONDARYHLAAAIAF

Single-Cell & Spatial Omics Discovery Platforms (target ID/validation, functional profiling)

Healthcare & Life Sciences/Biotechnology & Biologics/Biologic Drug Discovery Platforms (screening, antibody engineering, synthetic biology)

SECONDARYHLAAAIAI

Target Identification & Validation Platforms (omics-driven, chemoproteomics, genetic validation)

Healthcare & Life Sciences/Biotechnology & Biologics/Biologic Drug Discovery Platforms (screening, antibody engineering, synthetic biology)

Keywords

• Medical foundation models
• Multimodal clinical AI
• Pathology AI models
• Medical AI benchmarks
• Biomedical AI research

Headquarters

HQ city

Dover

HQ country

United States

HQ region

North America

Offices1 record

Markets served

GTM type

B2B

Offering type

Software

Cost components

Technology or R&D, Personnel, Infrastructure, Marketing or Sales, Operations

Revenue model

1. Enterprise Software Licenses/SaaS: Commercial products with pharma, hospitals, and medtech teams translating open-source models into clinically validated applications. First product in active development with clinical, biopharma, and life sciences partners.
2. API Access / Usage-based: Open-source models available publicly; commercial API access likely for enterprise applications requiring higher throughput or enterprise support.

Go-to-market motion3 records

Distribution channels2 records

Marketing channels6 records

Core offering

Sophont builds multimodal foundation models for medicine that unify clinical notes, labs, imaging (X-ray, CT, MRI), pathology, ECG/wearable signals, genomics, medications, and demographics into a single patient state representation. It releases open-source models (OpenMidnight for pathology, CortexMAE for fMRI), open benchmarks (Medmarks, Brainmarks), and runs the MedARC research community, while developing a first commercial product for clinical, biopharma, and life sciences partners.

Product overview

Sophont is an open medical superintelligence company developing multimodal foundation models for medicine. Their portfolio includes open-source pathology foundation models (OpenMidnight), functional MRI models (CortexMAE), medical LLM benchmarks (Medmarks), and fMRI benchmarks (Brainmarks), all part of their open science approach. They also run the nanopath research challenge and host the MedARC community of 4,000+ researchers. A commercial product translating these models into clinical applications is in development with biopharma and healthcare partners. All offerings are built around unifying clinical notes, labs, imaging, pathology, genomic data, and patient signals for disease trajectory prediction and intervention planning.

Differentiator

Problem solved

Functional benefit

Brands

• MedARC: The world's largest open medical AI research community, founded by Sophont's CEO, home to thousands of researchers, clinicians, and engineers with 4,000+ community members and 10 published papers.

Products and services

• OpenMidnight Open-source pathology foundation model for biomarker discovery and predicting clinical outcomes from histopathology slides (H&E, IHC); trained state-of-the-art with a $1.6k budget.
• CortexMAE Open-source functional MRI foundation model that improves signal quality and predicts clinical biomarkers of neurodegenerative and psychiatric disorders.
• Medmarks Open medical LLM benchmark suite evaluating clinically relevant capabilities with a live leaderboard; mentioned by OpenAI in the ChatGPT for Clinicians announcement.
• Brainmarks Open fMRI benchmark suite for predicting brain state and clinical diagnoses.
• nanopath Challenge Open challenge to train the best computational pathology foundation model in 1 hour on 1 GPU, advancing efficiency benchmarks for medical AI.
• MindEye2 Shared-subject fMRI-to-image reconstruction model requiring only 1 hour of data per subject; published at ICML 2024.
• Sophont Commercial Product (In Development) First commercial product translating Sophont's open-source foundation models into clinically validated applications for pharma, hospitals, and medtech partners; in active development.

Quantifiable outcome

• MindEye2 achieves fMRI-to-image reconstruction with only 1 hour of data (vs traditional approaches requiring much more)
• +2 more outcomes

Named customers2 records

Segments3 records

Ideal customer profiles3 records

Technology focussed Yes

API detail

Has API

No

API docs

API detail

Core technology

AI maturity

App detail

AI capability10 records

Feature8 records

Scale indicators3 records

Recent moves6 records

Expansion highlights6 records

Direct peers

• Owkin: Owkin applies AI to multimodal patient data (pathology, imaging, genomics) for biopharma and clinical research, including federated learning approaches. It directly overlaps Sophont's multimodal foundation-model thesis across pathology and biomedical research customers.
• Hippocratic AI: Hippocratic AI builds large language models targeted at healthcare workflows (clinical staffing, patient-facing assistants). It is a direct competitor to Sophont in applying foundation models to clinical use cases, with a similar enterprise and provider go-to-market motion.
• Recursion Pharmaceuticals: Recursion combines high-content biology data with foundation-model-style AI for drug discovery and translational research. It is comparable to Sophont's biopharma-oriented multimodal patient representation work and disease trajectory / intervention response themes.
• Paige.AI: Paige.AI builds AI for cancer pathology, including foundation models for slide analysis, and works with biopharma and clinical providers. It is one of the closest direct peers to Sophont's pathology foundation model work.
• PathAI: PathAI develops AI-powered pathology models for biomarker discovery and diagnostic support, including foundation-model efforts in computational pathology. It is a direct competitor to Sophont's OpenMidnight pathology foundation model and biopharma pathology customers.

Broad incumbents

• Tempus AI: Tempus operates a large multimodal clinical and molecular data platform with AI applications in oncology, cardiology, and neurology, and serves biopharma, payers, and providers. It is a broad incumbent whose scope overlaps Sophont's multimodal patient state vision, but at much greater scale.
• Google Health / Google DeepMind: Google's Med-PaLM / MedLM family and DeepMind health research target clinical LLMs and multimodal medical AI across providers and biopharma. It is a broad incumbent whose foundation-model ambitions and capital base dwarf Sophont's, while competing for the same enterprise customers.
• Microsoft Cloud for Healthcare / Nuance: Microsoft, through Nuance DAX Copilot and Cloud for Healthcare, ships medical AI ambient documentation and multimodal clinical intelligence into large EHR footprints. It is a broad incumbent competing for the same hospital and clinical-decision-support budgets Sophont targets.

Emerging players

• Mendel: Mendel builds AI to structure clinical and biomedical text and produce cohort-level analytics for life sciences and provider customers. It overlaps with Sophont's clinical-notes LLM capabilities and biopharma customer segment, but from a more text-centric starting point.
• Atropos Health: Atropos Health applies AI to real-world clinical data to generate evidence at the point of care for providers and biopharma. It overlaps with Sophont's evidence-generation and longitudinal patient representation themes while remaining a smaller, more focused emerging player.

Market position

Strengths4 records

Weaknesses4 records

Competitive moat6 records

Key risks6 records

Key highlights7 records

Customer concentration

• GitHubgithub.com/MedARC-AI
• LinkedInlinkedin.com/company/sophont
• Xx.com/SophontAI

Revenue estimate

Valuation estimate

Number of profiles

Profiles3 records

Funding overview

Funding rounds2 records

Investors8 records

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about Sophont